This month, 21 games are joining the cloud gaming library of over 2,000 titles. Whether chasing epic adventures, testing skills in competitive battles or diving into immersive worlds, members can dive into April’s adventures arrivals, which are truly no joke.
Get ready to stream, play and conquer the eight games available this week. Members can also get ahead of the pack with advanced access to South of Midnight, streaming soon before launch.
Unleash the Magic
South of Midnight, an action-adventure game developed by Compulsion Games, offers advanced access for gamers who purchase its Premium Edition. Dive into the title’s hauntingly beautiful world before launch, exploring its rich Southern gothic setting and unique magical combat system while balancing magic with melee attacks.
Step into the shadows.
Set in a mystical version of the American South, the game combines elements of magic, mystery and adventure, weaving a compelling story that draws players in. The endless opportunities for exploration and combat, along with deep lore and engaging characters, make the game a must-play for fans of the action-adventure genre.
With its blend of dark fantasy and historical influences, South of Midnight is poised to deliver a unique gaming experience that will leave players spellbound.
GeForce NOW members can be among the first to get advanced access to the game without the hassle of downloads or updates. With an Ultimate or Performance membership, experience the game’s haunting landscapes and cryptid encounters with the highest frame rates and lowest latency — no need for the latest hardware.
April Is Calling
Verdansk is back! Catch it in the cloud.
Verdansk, the original and iconic map from Call of Duty: Warzone, is making its highly anticipated return in the game’s third season, and available to stream on GeForce NOW. Known for its sprawling urban areas, rugged wilderness and points of interest like Dam and Superstore, Verdansk offers a dynamic battleground for intense combat. The map has been rebuilt from the ground up with key enhancements across audio, visuals and gameplay, getting back to basics and delivering nostalgia for fans.
Look for the following games available to stream in the cloud this week:
South of Midnight Advanced Access (Steam and Xbox, coming soon before launch)
AI is rapidly transforming how organizations solve complex challenges.
The early stages of enterprise AI adoption focused on using large language models to create chatbots. Now, enterprises are using agentic AI to create intelligent multi-agent systems that reason, act and execute complex tasks with a degree of autonomy.
Jacob Liberman, director of product management at NVIDIA, joined the NVIDIA AI Podcast to explain how agentic AI bridges the gap between powerful AI models and practical enterprise applications.
Enterprises are deploying AI agents to free human workers from time-consuming and error-prone tasks. This allows people to spend more time on high-value work that requires creativity and strategic thinking.
Liberman anticipates it won’t be long before teams of AI agents and human workers collaborate to tackle complex tasks requiring reasoning, intuition and judgement. For example, enterprise software developers will work with AI agents to develop more efficient algorithms. And medical researchers will collaborate with AI agents to design and test new drugs.
NVIDIA AI Blueprints help enterprises build their own AI agents – including many of the use cases listed above.
“Blueprints are reference architectures implemented in code that show you how to take NVIDIA software and apply it to some productive task in an enterprise to solve a real business problem,” Liberman said.
The blueprints are entirely open source. A developer or service provider can deploy a blueprint directly, or customize it by integrating their own technology.
Kanjun Qiu, CEO of Imbue, discusses the emerging era of personal AI agents, drawing a parallel to the PC revolution and explaining how modern AI systems are evolving to enhance user capabilities through collaboration.
Kaaren Hilsen, chief innovation officer and head of the AI factory at Telenor, highlights Norway’s first AI factory, which securely processes sensitive data within the country while promoting data sovereignty and environmental sustainability through green computing initiatives, including a renewable energy-powered data center in Oslo.
Jon Heller of Firsthand explains how the company’s AI Brand Agents are boosting retail and digital marketing by personalizing customer experiences and converting marketing interactions into valuable research data.
In the latest MLPerf Inference V5.0 benchmarks, which reflect some of the most challenging inference scenarios, the NVIDIA Blackwell platform set records — and marked NVIDIA’s first MLPerf submission using the NVIDIA GB200 NVL72 system, a rack-scale solution designed for AI reasoning.
Delivering on the promise of cutting-edge AI takes a new kind of compute infrastructure, called AI factories. Unlike traditional data centers, AI factories do more than store and process data — they manufacture intelligence at scale by transforming raw data into real-time insights. The goal for AI factories is simple: deliver accurate answers to queries quickly, at the lowest cost and to as many users as possible.
The complexity of pulling this off is significant and takes place behind the scenes. As AI models grow to billions and trillions of parameters to deliver smarter replies, the compute required to generate each token increases. This requirement reduces the number of tokens that an AI factory can generate and increases cost per token. Keeping inference throughput high and cost per token low requires rapid innovation across every layer of the technology stack, spanning silicon, network systems and software.
The latest updates to MLPerf Inference, a peer-reviewed industry benchmark of inference performance, include the addition of Llama 3.1 405B, one of the largest and most challenging-to-run open-weight models. The new Llama 2 70B Interactive benchmark features much stricter latency requirements compared with the original Llama 2 70B benchmark, better reflecting the constraints of production deployments in delivering the best possible user experiences.
In addition to the Blackwell platform, the NVIDIA Hopper platform demonstrated exceptional performance across the board, with performance increasing significantly over the last year on Llama 2 70B thanks to full-stack optimizations.
NVIDIA Blackwell Sets New Records
The GB200 NVL72 system — connecting 72 NVIDIA Blackwell GPUs to act as a single, massive GPU — delivered up to 30x higher throughput on the Llama 3.1 405B benchmark over the NVIDIA H200 NVL8 submission this round. This feat was achieved through more than triple the performance per GPU and a 9x larger NVIDIA NVLink interconnect domain.
While many companies run MLPerf benchmarks on their hardware to gauge performance, only NVIDIA and its partners submitted and published results on the Llama 3.1 405B benchmark.
Production inference deployments often have latency constraints on two key metrics. The first is time to first token (TTFT), or how long it takes for a user to begin seeing a response to a query given to a large language model. The second is time per output token (TPOT), or how quickly tokens are delivered to the user.
The new Llama 2 70B Interactive benchmark has a 5x shorter TPOT and 4.4x lower TTFT — modeling a more responsive user experience. On this test, NVIDIA’s submission using an NVIDIA DGX B200 system with eight Blackwell GPUs tripled performance over using eight NVIDIA H200 GPUs, setting a high bar for this more challenging version of the Llama 2 70B benchmark.
Combining the Blackwell architecture and its optimized software stack delivers new levels of inference performance, paving the way for AI factories to deliver higher intelligence, increased throughput and faster token rates.
NVIDIA Hopper AI Factory Value Continues Increasing
The NVIDIA Hopper architecture, introduced in 2022, powers many of today’s AI inference factories, and continues to power model training. Through ongoing software optimization, NVIDIA increases the throughput of Hopper-based AI factories, leading to greater value.
On the Llama 2 70B benchmark, first introduced a year ago in MLPerf Inference v4.0, H100 GPU throughput has increased by 1.5x. The H200 GPU, based on the same Hopper GPU architecture with larger and faster GPU memory, extends that increase to 1.6x.
Hopper also ran every benchmark, including the newly added Llama 3.1 405B, Llama 2 70B Interactive and graph neural network tests. This versatility means Hopper can run a wide range of workloads and keep pace as models and usage scenarios grow more challenging.
It Takes an Ecosystem
This MLPerf round, 15 partners submitted stellar results on the NVIDIA platform, including ASUS, Cisco, CoreWeave, Dell Technologies, Fujitsu, Giga Computing, Google Cloud, Hewlett Packard Enterprise, Lambda, Lenovo, Oracle Cloud Infrastructure, Quanta Cloud Technology, Supermicro, Sustainable Metal Cloud and VMware.
The breadth of submissions reflects the reach of the NVIDIA platform, which is available across all cloud service providers and server makers worldwide.
MLCommons’ work to continuously evolve the MLPerf Inference benchmark suite to keep pace with the latest AI developments and provide the ecosystem with rigorous, peer-reviewed performance data is vital to helping IT decision makers select optimal AI infrastructure.
The 4:2:2 color format is a game changer for professional video editors, as it retains nearly as much color information as 4:4:4 while greatly reducing file size. This improves color grading and chroma keying — using color information to isolate a specific range of hues — while maximizing efficiency and quality.
Adobe and other industry partners are attending NAB Show — a premier gathering of over 100,000 leaders in the broadcast, media and entertainment industries — running April 5-9 in Las Vegas. Professionals in these fields will come together for education, networking and exploring the latest technologies and trends.
Shed Some Color on 4:2:2
Consumer cameras that are limited to 4:2:0 color compression capture a limited amount of color information. 4:2:0 is acceptable for video playback on browsers, but professional video editors often rely on cameras that capture 4:2:2 color depth with precise color accuracy to ensure higher color fidelity.
Adobe Premiere Pro’s beta with 4:2:2 means video data can now provide double the color information with just a 1.3x increase in raw file size over 4:2:0. This unlocks several key benefits within professional video-production workflows:
Increased Color Accuracy: 10-bit 4:2:2 retains more color information compared with 8-bit 4:2:0, leading to more accurate color representation and better color grading results.
4:2:2 offers more accurate color representation for better color grading results.
More Flexibility: The extra color data allows for increased flexibility during color correction and grading, enabling more nuanced adjustments and corrections.
Improved Keying: 4:2:2 is particularly beneficial for keying — including green screening — as it enables cleaner, more accurate extraction of the subject from the background, as well as cleaner edges of small keyed objects like hair.
4:2:2 enables cleaner green screen video content.
Smaller File Sizes: Compared with 4:4:4, 4:2:2 reduces file sizes without significantly impacting picture quality, offering an optimal balance between quality and storage.
Combining 4:2:2 support with NVIDIA hardware increases creative possibilities.
Advanced Video Editing
Prosumer-grade cameras from most major brands support HEVC and H.264 10-bit 4:2:2 formats to deliver superior image quality, manageable file sizes and the flexibility needed for professional video production.
GeForce RTX 50 Series GPUs paired with Microsoft Windows 11 come with GPU-powered decode acceleration in HEVC and H.264 10-bit 4:2:2 formats.
GPU-powered decode enables faster-than-real-time playback without stuttering, the ability to work with original camera media instead of proxies, smoother timeline responsiveness and reduced CPU load — freeing system resources for multi-app workflows and creative tasks.
RTX 50 Series’ 4:2:2 hardware can decode up to six 4K 60 frames-per-second video sources on an RTX 5090-enabled studio PC, enabling smooth multi-camera video-editing workflows on Adobe Premiere Pro.
Video exports are also accelerated with NVIDIA’s ninth-generation encoder and sixth-generation decoder.
NVIDIA and GeForce RTX Laptop GPU encoders and decoders.
In GeForce RTX 50 Series GPUs, the ninth-generation NVIDIA video encoder, NVENC, offers an 8% BD-BR upgrade in video encoding efficiency when exporting to HEVC on Premiere Pro.
Adobe AI Accelerated
Adobe delivers an impressive array of advanced AI features for idea generation, enabling streamlined processes, improved productivity and opportunities to explore new artistic avenues — all accelerated by NVIDIA RTX GPUs.
For example, Adobe Media Intelligence, a feature in Premiere Pro (beta) and After Effects (beta), uses AI to analyze footage and apply semantic tags to clips. This lets users more easily and quickly find specific footage by describing its content, including objects, locations, camera angles and even transcribed spoken words.
Media Intelligence runs 30% faster on the GeForce RTX 5090 Laptop GPU compared with the GeForce RTX 4090 Laptop GPU.
In addition, the Enhance Speech feature in Premiere Pro (beta) improves the quality of recorded speech by filtering out unwanted noise and making the audio sound clearer and more professional. Enhance Speech runs 7x faster on GeForce RTX 5090 Laptop GPUs compared to the MacBook Pro M4 Max.
Visit Adobe’s Premiere Pro page to download a free trial of the beta and explore the slew of AI-powered features across the Adobe Creative Cloud and Substance 3D apps.
Unleash (AI)nfinite Possibilities
GeForce RTX 5090 and 5080 Series laptops deliver the largest-ever generational leap in portable performance for creating, gaming and all things AI.
They can run creative generative AI models such as Flux up to 2x faster in a smaller memory footprint, compared with the previous generation.
The previously mentioned ninth-generation NVIDIA encoders elevate video editing and livestreaming workflows, and come with NVIDIA DLSS 4 technology and up to 24GB of VRAM to tackle massive 3D projects.
NVIDIA Max-Q hardware technologies use AI to optimize every aspect of a laptop — the GPU, CPU, memory, thermals, software, display and more — to deliver incredible performance and battery life in thin and quiet devices.
All GeForce RTX 50 Series laptops include NVIDIA Studio platform optimizations, with over 130 GPU-accelerated content creation apps and exclusive Studio tools including NVIDIA Studio Drivers, tested extensively to enhance performance and maximize stability in popular creative apps.
The game-changing NVIDIA GeForce RTX 5090 and 5080 GPU laptops are available now.
Adobe will participate in the Creator Lab at NAB Show, offering hands-on training for editors to elevate their skills with Adobe tools. Attend a 30-minute section and try out Puget Systems laptops equipped with GeForce RTX 5080 Laptop GPUs to experience blazing-fast performance and demo new generative AI features.
Advances in physical AI are enabling organizations to embrace embodied AI across their operations, bringing unprecedented intelligence, automation and productivity to the world’s factories, warehouses and industrial facilities.
Humanoid robots can work alongside human teams, autonomous mobile robots (AMRs) can navigate complex warehouse environments, and intelligent cameras and visual AI agents can monitor and optimize entire facilities. In these ways, physical AI is becoming integral to today’s industrial operations.
At Hannover Messe — a trade show on industrial development running through April 4 in Germany — manufacturing, warehousing and supply chain leaders such as Accenture and Schaeffler are showcasing their adoption of the blueprint to simulate Digit, a humanoid robot from Agility Robotics, and discussing how they use industrial AI and digital twins to optimize facility layouts, material flow and collaboration between humans and robots inside complex production environments.
Digital Twins — the Training Ground for Physical AI
Industrial facility digital twins are physically accurate virtual replicas of real-world facilities that serve as critical testing grounds for simulating and validating physical AI and how robots and autonomous fleets interact, collaborate and tackle complex tasks before deployment.
Developers can use NVIDIA Omniverse platform technologies and the Universal Scene Description (OpenUSD) framework to develop digital twins of their facilities and processes. This simulation-first approach dramatically accelerates development cycles while reducing the costs and risks associated with real-world testing.
Built for a Diversity of Robots and AI Agents
The Mega blueprint equips industrial enterprises with a reference workflow for combining sensor simulation and synthetic data generation to simulate complex human-robot interactions and verify the performance of autonomous systems in industrial digital twins.
Enterprises can use Mega to test various robot brains and policies at scale for mobility, navigation, dexterity and spatial reasoning. This enables fleets comprising different types of robots to work together as a coordinated system.
As robot brains execute their missions in simulation, they perceive the results of their actions through sensor simulation and plan their next action. This cycle continues until the policies are refined and ready for deployment.
Once validated, these policies are deployed to real robots, which continue to learn from their environment — sending sensor information back through the entire loop and creating a continuous learning and improvement cycle.
Transforming Industrial Operations With Visual AI Agents
In addition to AMRs and humanoid robots, advanced visual AI agents extract information from live and recorded video data, enabling new levels of intelligence and automation. These visual AI agents bring real-time contextual awareness to robots and help to improve worker safety, maintain warehouse compliance, support visual inspection and maximize space utilization.
To support developers building visual AI agents, which can be integrated with the Mega blueprint, NVIDIA last year announced an AI Blueprint for video search and summarization (VSS). At Hannover Messe, leading partners are featuring how they use the VSS blueprint to improve productivity and operational efficiency.
Accelerating Industrial Digitalization
The industrial world is now experiencing its software-defined moment, with visual AI agents and digital twins as the training ground for physical AI.
Explore the new self-paced Learn OpenUSD training curriculum that includes free NVIDIA Deep Learning Institute courses for 3D practitioners and developers.
Featured image courtesy of Accenture, Agility Robotics and Schaeffler.
See notice regarding software product information.
A new resident is moving into the cloud — KRAFTON’s inZOI joins the 2,000+ games in the GeForce NOW cloud gaming library.
Plus, members can get ready for an exclusive sneak peek as the Sunderfolk First Look Demo comes to the cloud. The demo is exclusively available for players on GeForce NOW until April 7, including Performance and Ultimate members as well as free users.
And explore the world of Atomfall — part of 12 games joining the cloud this week.
Cloud of Possibilities
Live the life of your dreams in the cloud.
In inZOI — a groundbreaking life simulation game by Krafton that pushes the genre’s boundaries — take on the role of an intern at AR COMPANY, managing virtual beings called “Zois” in a simulated city.
The game features over 400 mental elements influencing Zois’ behaviors. Experience the game’s dynamic weather system, open-world environments inspired by real locations and cinematic cut scenes for key life events — and even create in-game objects. inZOI lets players craft unique stories and live out their dreams in a meticulously designed virtual world.
Dive into the world of Zois without the need for high-end hardware. Members can manage their virtual homes, customize characters and explore the game’s dynamic environments from various devices, streaming its detailed graphics and complex simulations with ease.
A Magical Gateway
Sunderfolk’s First Look Demo has arrived on GeForce NOW, offering a tantalizing look into the magical realm of the Sunderlands. Designed as a TV-first experience, this shared-turn-based tactical role-playing game (RPG) enables using a mobile phone as the gameplay controller. Up to four players can gather around the big screen and embark on a journey filled with strategic battles.
This second-screen approach keeps players engaged in real time, adding new layers of immersion. With all six unique character classes unlocked from the start, players can experience the early hours of the game, experimenting with different team compositions and tactics to overcome the challenges that await.
Let the magic begin.
Accessing the demo is a breeze — head to the GeForce NOW app, select Sunderfolk and jump right in. Explore the Sunderlands, engage in flexible turn-based combat and help rebuild the village of Arden to get a taste of the full game’s depth and camaraderie.
Gather the gaming squad, grab a phone and prepare to write a completely new legend in this RPG adventure. The First Look Demo is only available on GeForce NOW, where members can enjoy high-quality graphics and seamless gameplay on their phones and tablets, along with the innovative mobile-as-controller mechanic that makes Sunderfolk’s couch co-op experience so engaging.
Epic Adventures Await
Enter a world where danger lurks in every shadow.
Blending folk horror and intense combat, Atomfall is a survival-action game set in an alternate 1960s Britain, where the Windscale nuclear disaster has left Northern England a radioactive wasteland. Players explore eerie open zones filled with mutated creatures, cultists and Cold War mysteries while scavenging resources, crafting weapons and uncovering the truth behind the disaster. GeForce NOW members can stream it today across their devices of choice.
Look for the following games available to stream in the cloud this week:
Sunderfolk First Look Demo (New release, March 25)
Atomfall (New release on Steam and Xbox available on PC Game Pass, March 27)
The First Berserker: Khazan (New release on Steam, March 27)
From handling demand surges and evolving power needs to preventing infrastructure failures that can cause wildfires, utility companies have a lot to keep tabs on.
Buzz Solutions — a member of the NVIDIA Inception program for cutting-edge startups — is helping by using AI to improve how utilities monitor and maintain their infrastructure.
Kaitlyn Albertoli, CEO and cofounder of Buzz Solutions joined the AI Podcast to explain how the company’s vision AI technology helps utilities spot potential problems faster.
Buzz Solutions helps utility companies analyze the massive amounts of inspection data collected by drones and helicopters. The company’s proprietary machine learning algorithms identify potential issues ranging from broken and rusted components to encroaching vegetation and unwelcome wildlife visits — before they cause outages or wildfires.
To help address substation issues, Buzz Solutions built PowerGUARD, a container-based application pipeline that uses AI to analyze video streams from substation cameras in real time. It detects security, safety, fire, smoke and equipment issues, annotates the video, then sends alerts via email or to a dashboard.
PowerGUARD uses the NVIDIA DeepStream software development kit for processing and inference of video streams used in real-time video analytics. DeepStream runs within the NVIDIA Metropolis framework on the NVIDIA Jetson edge AI platform or on cloud-based virtual machines to improve performance, reduce costs and save time.
Albertoli believes AI is just getting started in the utility industry, as it enables workers to take action rather than spend months reviewing images manually. “We are just at the tip of the iceberg of seeing AI enter into the energy sector and start to provide real value,” she said.
Time Stamps
05:15: How Buzz Solutions saw an opportunity in the massive amounts of inspection data utility companies were collecting but not analyzing.
12:25: The importance of modernizing energy infrastructure with actionable intelligence.
16:27: How AI identifies critical risks like rusted components, vegetation encroachment and sparking issues before they cause wildfires.
20:00: Buzz Solutions’ innovative use of synthetic data to train algorithms for rare events.
Telenor opened Norway’s first AI factory in November 2024, enabling organizations to process sensitive data securely on Norwegian soil while prioritizing environmental responsibility. Telenor’s Chief Innovation Officer and Head of the AI Factory Kaaren Hilsen discusses the AI factory’s rapid development, going from concept to reality in under a year.
AI isn’t just about building smarter machines. It’s about building a greener world. AI and accelerated computing are helping industries tackle some of the world’s toughest environmental challenges. Joshua Parker, senior director of corporate sustainability at NVIDIA, explains how these technologies are powering a new era of energy efficiency.
AI is everywhere. So, too, are concerns about advanced technology’s environmental impact. Daniel Castro, vice president of the Information Technology and Innovation Foundation and director of its Center for Data Innovation, discusses his AI energy use report that addresses misconceptions about AI’s energy consumption. He also talks about the need for continued development of energy-efficient technology.
Generative AI is unlocking new capabilities for PCs and workstations, including game assistants, enhanced content-creation and productivity tools and more.
NVIDIA NIM microservices, available now, and AI Blueprints, coming in April, accelerate AI development and improve its accessibility. Announced at the CES trade show in January, NVIDIA NIM provides prepackaged, state-of-the-art AI models optimized for the NVIDIA RTX platform, including the NVIDIA GeForce RTX 50 Series and, now, the new NVIDIA Blackwell RTX PRO GPUs. The microservices are easy to download and run. They span the top modalities for PC development and are compatible with top ecosystem applications and tools.
The experimental System Assistant feature of Project G-Assist was also released today. Project G-Assist showcases how AI assistants can enhance apps and games. The System Assistant allows users to run real-time diagnostics, get recommendations on performance optimizations, or control system software and peripherals — all via simple voice or text commands. Developers and enthusiasts can extend its capabilities with a simple plug-in architecture and new plug-in builder.
Amid a pivotal moment in computing — where groundbreaking AI models and a global developer community are driving an explosion in AI-powered tools and workflows — NIM microservices, AI Blueprints and G-Assist are helping bring key innovations to PCs. This RTX AI Garage blog series will continue to deliver updates, insights and resources to help developers and enthusiasts build the next wave of AI on RTX AI PCs and workstations.
Ready, Set, NIM!
Though the pace of innovation with AI is incredible, it can still be difficult for the PC developer community to get started with the technology.
Bringing AI models from research to the PC requires curation of model variants, adaptation to manage all of the input and output data, and quantization to optimize resource usage. In addition, models must be converted to work with optimized inference backend software and connected to new AI application programming interfaces (APIs). This takes substantial effort, which can slow AI adoption.
NVIDIA NIM microservices help solve this issue by providing prepackaged, optimized, easily downloadable AI models that connect to industry-standard APIs. They’re optimized for performance on RTX AI PCs and workstations, and include the top AI models from the community, as well as models developed by NVIDIA.
NIM microservices support a range of AI applications, including large language models (LLMs), vision language models, image generation, speech processing, retrieval-augmented generation (RAG)-based search, PDF extraction and computer vision. Ten NIM microservices for RTX are available, supporting a range of applications, including language and image generation, computer vision, speech AI and more. Get started with these NIM microservices today:
NIM microservices are also available through top AI ecosystem tools and frameworks.
For AI enthusiasts, AnythingLLM and ChatRTX now support NIM, making it easy to chat with LLMs and AI agents through a simple, user-friendly interface. With these tools, users can create personalized AI assistants and integrate their own documents and data, helping automate tasks and enhance productivity.
For developers looking to build, test and integrate AI into their applications, FlowiseAI and Langflow now support NIM and offer low- and no-code solutions with visual interfaces to design AI workflows with minimal coding expertise. Support for ComfyUI is coming soon. With these tools, developers can easily create complex AI applications like chatbots, image generators and data analysis systems.
In addition, Microsoft VS Code AI Toolkit, CrewAI and Langchain now support NIM and provide advanced capabilities for integrating the microservices into application code, helping ensure seamless integration and optimization.
NVIDIA AI Blueprints Will Offer Pre-Built Workflows
NVIDIA AI Blueprints, coming in April, give AI developers a head start in building generative AI workflows with NVIDIA NIM microservices.
Blueprints are ready-to-use, extensible reference samples that bundle everything needed — source code, sample data, documentation and a demo app — to create and customize advanced AI workflows that run locally. Developers can modify and extend AI Blueprints to tweak their behavior, use different models or implement completely new functionality.
PDF to podcast AI Blueprint coming soon.
The PDF to podcast AI Blueprint will transform documents into audio content so users can learn on the go. By extracting text, images and tables from a PDF, the workflow uses AI to generate an informative podcast. For deeper dives into topics, users can then have an interactive discussion with the AI-powered podcast hosts.
The AI Blueprint for 3D-guided generative AI will give artists finer control over image generation. While AI can generate amazing images from simple text prompts, controlling image composition using only words can be challenging. With this blueprint, creators can use simple 3D objects laid out in a 3D renderer like Blender to guide AI image generation. The artist can create 3D assets by hand or generate them using AI, place them in the scene and set the 3D viewport camera. Then, a prepackaged workflow powered by the FLUX NIM microservice will use the current composition to generate high-quality images that match the 3D scene.
NVIDIA NIM on RTX With Windows Subsystem for Linux
One of the key technologies that enables NIM microservices to run on PCs is Windows Subsystem for Linux (WSL).
Microsoft and NVIDIA collaborated to bring CUDA and RTX acceleration to WSL, making it possible to run optimized, containerized microservices on Windows. This allows the same NIM microservice to run anywhere, from PCs and workstations to the data center and cloud.
Project G-Assist Expands PC AI Features With Custom Plug-Ins
As part of Project G-Assist, an experimental version of the System Assistant feature for GeForce RTX desktop users is now available via the NVIDIA App, with laptop support coming soon.
G-Assist helps users control a broad range of PC settings — including optimizing game and system settings, charting frame rates and other key performance statistics, and controlling select peripherals settings such as lighting — all via basic voice or text commands.
G-Assist is built on NVIDIA ACE — the same AI technology suite game developers use to breathe life into non-player characters. Unlike AI tools that use massive cloud-hosted AI models that require online access and paid subscriptions, G-Assist runs locally on a GeForce RTX GPU. This means it’s responsive, free and can run without an internet connection. Manufacturers and software providers are already using ACE to create custom AI Assistants like G-Assist, including MSI’s AI Robot engine, the Streamlabs Intelligent AI Assistant and upcoming capabilities in HP’s Omen Gaming hub.
G-Assist was built for community-driven expansion. Get started with this NVIDIA GitHub repository, including samples and instructions for creating plug-ins that add new functionality. Developers can define functions in simple JSON formats and drop configuration files into a designated directory, allowing G-Assist to automatically load and interpret them. Developers can even submit plug-ins to NVIDIA for review and potential inclusion.
Currently available sample plug-ins include Spotify, to enable hands-free music and volume control, and Google Gemini — allowing G-Assist to invoke a much larger cloud-based AI for more complex conversations, brainstorming sessions and web searches using a free Google AI Studio API key.
In the clip below, you’ll see G-Assist ask Gemini about which Legend to pick in Apex Legends when solo queueing, and whether it’s wise to jump into Nightmare mode at level 25 in Diablo IV:
For even more customization, follow the instructions in the GitHub repository to generate G-Assist plug-ins using a ChatGPT-based “Plug-in Builder.” With this tool, users can write and export code, then integrate it into G-Assist — enabling quick, AI-assisted functionality that responds to text and voice commands.
Watch how a developer used the Plug-in Builder to create a Twitch plug-in for G-Assist to check if a streamer is live:
Check out the G-Assist article for system requirements and additional information.
Build, Create, Innovate
NVIDIA NIM microservices for RTX are available at build.nvidia.com, providing developers and AI enthusiasts with powerful, ready-to-use tools for building AI applications.
Download Project G-Assist through the NVIDIA App’s “Home” tab, in the “Discovery” section. G-Assist currently supports GeForce RTX desktop GPUs, as well as a variety of voice and text commands in the English language. Future updates will add support for GeForce RTX Laptop GPUs, new and enhanced G-Assist capabilities, as well as support for additional languages. Press “Alt+G” after installation to activate G-Assist.
Each week, RTX AI Garage features community-driven AI innovations and content for those looking to learn more about NIM microservices and AI Blueprints, as well as building AI agents, creative workflows, digital humans, productivity apps and more on AI PCs and workstations.
Time to sharpen the blade. GeForce NOW brings a legendary addition to the cloud: Ubisoft’s highly anticipated Assassin’s Creed Shadows is now available for members to stream.
Plus, dive into the updated version of the iconic Fable Anniversary — part of 11 games joining the cloud this week.
Silent as a Shadow
Take the Leap of Faith from the cloud.
Explore 16th-century Japan, uncover conspiracies and shape the destiny of a nation — all from the cloud.
Assassin’s Creed Shadows unfolds in 1579, during the turbulent Azuchi-Momoyama period of feudal Japan, a time of civil war and cultural exchange.
Step into the roles of Naoe, a fictional shinobi assassin and daughter of Fujibayashi Nagato, and Yasuke, a character based on the historical African samurai. Their stories intertwine as they find themselves on opposite sides of a conflict.
The game’s dynamic stealth system enables players to hide in shadows and use a new “Observe” mechanic to identify targets, tag enemies and highlight objectives. Yasuke and Naoe each have unique abilities and playstyles: Naoe excels in stealth, equipped with classic Assassin techniques and shinobi skills, while Yasuke offers a more combat-focused approach.
Navigate the turbulent Sengoku period on GeForce NOW, and experience the game’s breathtaking landscapes and intense combat at up to 4K resolution and 120 frames per second with an Ultimate membership. Every sword clash and sweeping vista is delivered with exceptional smoothness and clarity.
A Classic Reborn
Fable Anniversary revitalizes the original Fable: The Lost Chapters with enhanced graphics, a new save system and Xbox achievements. This action role-playing game invites players to shape their heroes’ destinies in the whimsical world of Albion.
Make every choice from the cloud.
Fable Anniversary weaves an epic tale of destiny and choice, following the journey of a young boy whose life is forever changed when bandits raid his peaceful village of Oakvale. Recruited to the Heroes’ Guild, he embarks on a quest to uncover the truth about his family and confront the mysterious Jack of Blades.
Players shape their hero’s destiny through a series of moral choices. These decisions influence the story’s progression and even manifest physically on the character.
Stream the title with a GeForce NOW membership across PCs that may not be game-ready, Macs, mobile devices, and Samsung and LG smart TVs. GeForce NOW transforms these devices into powerful gaming rigs, with up to eight-hour gaming sessions for Ultimate members.
Unleash the Games
Crash, smash, repeat.
Wreckfest 2, the highly anticipated sequel by Bugbear Entertainment to the original demolition derby racing game, promises an even more intense and chaotic experience. The game features a range of customizable cars, from muscle cars to novelty vehicles, each with a story to tell.
Play around with multiple modes, including traditional racing with physics-driven handling, and explore demolition derby arenas where the goal is to cause maximum destruction. With enhanced multiplayer features, including skills-based matchmaking and split-screen mode, Wreckfest 2 is the ultimate playground for destruction-racing enthusiasts.
Look for the following games available to stream in the cloud this week:
The power and utilities sector keeps the lights on for the world’s populations and industries. As the global energy landscape evolves, so must the tools it relies on.
To advance the next generation of electricity generation and distribution, many of the industry’s members are joining forces through the creation of the Open Power AI Consortium. The consortium includes energy companies, technology companies and researchers developing AI applications to tackle domain-specific challenges, such as adapting to an increased deployment of distributed energy resources and significant load growth on electric grids.
Led by independent, nonprofit energy R&D organization EPRI, the consortium aims to spur AI adoption in the power sector through a collaborative effort to build open models using curated, industry-specific data. The initiative was launched today at NVIDIA GTC, a global AI conference taking place through Friday, March 21, in San Jose, California.
“Over the next decade, AI has the great potential to revolutionize the power sector by delivering the capability to enhance grid reliability, optimize asset performance, and enable more efficient energy management,” said Arshad Mansoor, EPRI’s president and CEO. “With the Open Power AI Consortium, EPRI and its collaborators will lead this transformation, driving innovation toward a more resilient and affordable energy future.”
As part of the consortium, EPRI, NVIDIA and Articul8, a member of the NVIDIA Inception program for cutting-edge startups, are developing a set of domain-specific, multimodal large language models trained on massive libraries of proprietary energy and electrical engineering data from EPRI that can help utilities streamline operations, boost energy efficiency and improve grid resiliency.
The first version of an industry-first open AI model for electric and power systems was developed using hundreds of NVIDIA H100 GPUs and is expected to soon be available in early access as an NVIDIA NIM microservice.
“Working with EPRI, we aim to leverage advanced AI tools to address today’s unique industry challenges, positioning us at the forefront of innovation and operational excellence,” said Vincent Sorgi, CEO of PPL Corporation and EPRI board chair.
PPL is a leading U.S. energy company that provides electricity and natural gas to more than 3.6 million customers in Pennsylvania, Kentucky, Rhode Island and Virginia.
The Open AI Consortium’s Executive Advisory Committee includes executives from over 20 energy companies such as Duke Energy, Pacific Gas & Electric Company and Portland General Electric, as well as leading tech companies such as AWS, Oracle and Microsoft. The consortium plans to further expand its global member base.
Powering Up AI to Energize Operations, Drive Innovation
Global energy consumption is projected to grow by nearly 4% annually through 2027, according to the International Energy Agency. To support this surge in demand, electricity providers are looking to enhance the resiliency of power infrastructure, balance diverse energy sources and expand the grid’s capacity.
AI agents trained on thousands of documents specific to this sector — including academic research, industry regulations and standards, and technical documents — can enable utility and energy companies to more quickly assess energy needs and prepare the studies and permits required to improve infrastructure.
“We can bring AI to the global power sector in a much more accelerated way by working together to develop foundation models for the industry, and collaborating with the power sector to y apply solutions tailored to its unique needs,” Mansoor said.
Utilities could tap the consortium’s model to help accelerate interconnection studies, which analyze the feasibility and potential impact of connecting new generators to the existing electric grid. The process varies by region but can take up to four years to complete. By introducing AI agents that can support the analysis, the consortium aims to cut this timeline down by at least 5x.
The AI model could also be used to support the preparation of licenses, permits, environmental studies and utility rate cases, where energy companies seek regulatory approval and public comment on proposed changes to electricity rates.
Beyond releasing datasets and models, the consortium also aims to develop a standardized framework of benchmarks to help utilities, researchers and other energy sector stakeholders evaluate the performance and reliability of AI technologies.
Learn more about the Open Power AI Consortium online and in EPRI’s sessions at GTC:
NVIDIA GeForce NOW (@NVIDIAGFN) April 2, 2025
